Should you believe what you read? Reliability and validity in published pediatric nursing research.

A S MORE RESEARCH is conducted and disseminated, either through publication or other means, nurses are examining the findings of these studies to determine which have implications for their clinical practice. As alluring as findings or results can sometimes be, findings are meaningless without an evaluation of other aspects of the research. Findings from poorly designed or implemented studies are not only useless but they may also be dangerous, since if they are taken at face value, they can place nurses and patients in harmful or threatening situations. All aspects of a research study should be evaluated before decisions are made about the quality of the findings; no aspect is more important than the quality of the instrumentation used. This column is a review of some of the aspects of instrumentation in recently published pediatric nursing research. First, two psychometric concepts, reliability and validity, need to be briefly discussed. Reliability is the ratio of desired information to all obtained information (what you wanted compared with what you got) (Cronbach, 1970) or the extent to which measurement error is minimized (Nunnally, 1978; Polit & Hungler, 1983). To visualize this, imagine that you intended to determine the pulses of 10 children and then did so, completely accurately. You would have accurately measured each child’s pulse and your reliability as a “pulse instrument” would be perfect or 1 .OO. Reliability can range from 1.00 to + 1 .OO and is considered minimally acceptable at a value of 0.70 (Polit & Hungler, 1983) There are several types of reliability, with the most common ones being internal consistency (assessing for the consistency in a person’s response patterns) or test-retest (the consistency or stability of a person’s responses over time). Reliability is